Call Centers Hit the Right Note with Real-Time Accent Tweaks by AI

In a move that has kicked off a debate around the role of AI in global business communication, Teleperformance, the world’s largest call centre operator, has begun using artificial intelligence to modify Indian accents for western customers in real time. 

The French company claims the technology, developed by US-based startup Sanas, helps reduce miscommunication, increase productivity, and even protect call centre workers from verbal abuse.

However, the question is whether to see this AI-driven accent modification as a breakthrough in customer service or if it raises ethical concerns about linguistic identity and bias.

This was coming. Also the fake call centers duping western citizens for millions didn't help either. Big crash coming.

— Doesn't Matter ☣ (@KafarkPadtaHai) March 3, 2025

With a workforce of around 90,000 employees in India and tens of thousands more worldwide, Teleperformance plays a crucial role in serving UK customers. Its clients range from major corporations like Vodafone and eBay to essential public services, including the government and the NHS.

Case for Accent Neutralisation

Call centres have long been a backbone of global customer service, with India emerging as a major hub due to its large English-speaking workforce. However, customers, particularly in western markets, sometimes struggle to understand foreign accents, leading to frustration, longer call times, and increased escalations. 

In an exclusive conversation with AIM, Apurv Agrawal, CEO and co-founder of SquadStack, explained: “Companies are responding to customers’ expectations for seamless communication. Our research at SquadStack shows that communication barriers can reduce conversion rates by up to 23%. 

“While customer demand is a driving factor, operational benefits are equally compelling. These systems can reduce call handling times by 15-20% and minimise the need for repetition, which directly impacts efficiency.”

Agrawal noted that AI accent modification allows businesses to tap into a global talent pool without requiring extensive language training. 

“Rather than forcing agents to adopt unnatural accents through traditional training, AI allows them to communicate authentically while ensuring clarity.”

The Tech Behind Accent Modification

The technology behind real-time accent modification is sophisticated. AI models first use automatic speech recognition (ASR) to transcribe spoken words, followed by natural language processing (NLP) to identify phonetic patterns, stress variations, and intonations unique to the speaker’s accent. 

Transformation algorithms then adjust these elements to align with more widely understood pronunciation norms while retaining the speaker’s natural tone and emotion.

And the result is speech that sounds more “neutral” to the listener, delivered with minimal latency, often under 100 milliseconds.

According to Agrawal, the best AI accent-modification systems use multimodal Transformer architectures that process both acoustic and linguistic information simultaneously. “These systems ensure that conversations remain fluid, reducing misunderstandings and enhancing customer satisfaction.”

AI for Accent Modification is a Two-Way Street?

While companies like Teleperformance use AI to make Indian agents sound more “western”, some firms are exploring the reverse, helping Indian customers better understand foreign call centre agents.

SquadStack, for instance, has developed AI-driven communication enhancement tools that work bi-directionally, ensuring mutual comprehension rather than erasing accents.

“Our technology doesn’t eliminate accents but enhances clarity by adjusting speech patterns that typically cause misunderstandings. We’ve found this approach respects cultural identities while improving comprehension by approximately 40%,” said Agrawal.

Convenience or Ethics or Scam 

While AI-powered accent modification solves real business challenges, it also raises important ethical questions. Critics argue that “whitening” accents reinforce linguistic biases and perpetuate a hierarchy of acceptable English pronunciations. 

Teleperformance defends the move, stating that the AI simply removes barriers to understanding, improving efficiency without storing or altering data long-term. However, the debate continues: Should businesses prioritise convenience over cultural authenticity, or is there a way to balance both?

On the downside, AI-driven accent modification in call centres could make scam calls more convincing and harder to detect. Scammers often rely on social engineering, and a foreign accent has traditionally been a red flag. 

With AI neutralising accents, fraudsters can impersonate banks, tech support, or government agencies more convincingly, making scams more effective. While AI transforms customer service, it’s crucial to ensure it doesn’t empower scammers.